ai-executive-assistant

How connecting your tools via MCP transforms an AI agent from a smart guesser into an informed operator — and why it's simpler than you think.

What I learned building my first AI agent from scratch — what surprised me, what I got wrong, and what I'd tell someone starting today.

How to turn Claude Code into a persistent agent using the three-layer pattern — skills for capabilities, a profile for identity, and context files for memory across sessions.

A four-phase daily workflow — morning planning, active work, afternoon review, and nightly cleanup — where each phase feeds into the next so you never start from zero.

A prompt framework that builds your daily plan around energy levels — matching deep work to peak focus, admin to low-energy windows, with an 80% capacity buffer to prevent overcommitting.

A structured prompt that generates a pre-meeting brief — who you're meeting, why it matters, what's open, and the questions you should ask.

A simple framework for sizing tasks, setting daily capacity, and applying the 80% buffer rule so you stop overcommitting and start finishing what you planned.

A data-driven weekly retrospective prompt that calculates completion rate, overcommit score, and 3-week rolling trends to surface productivity patterns you can't see in the moment.

A terminal-based AI executive assistant built on Claude Code that manages your day, tracks tasks, preps meetings, and keeps you accountable.